KCNA: Kubernetes and machine learning (ML) workload

Question

In a Kubernetes cluster, you need to ensure that a specific machine learning (ML) workload, which requires GPUs for accelerated processing, is scheduled only on nodes equipped with the necessary GPU resources. Which of the following approaches would most effectively (without unnecessary complexity) achieve this while maintaining flexibility and scalability in the cluster?

quizzes.technology · Accepted Answer

Implementing a nodeSelector in the pod spec to restrict scheduling to nodes labeled with the required GPU resources

quizzes.technology · Answer

Specifying the nodeName field in the pod spec to directly assign the pod to a specific node

quizzes.technology · Answer

Using pod topology spread constraints to distribute the pods evenly across nodes with GPUs

quizzes.technology · Answer

Applying affinity and anti-affinity rules to schedule the pod on nodes labeled with specific GPU types

☁️ Free practice test question from: KCNA (KCNA)

Question: 269

Question: 323

Question: 362

💬 Discussion and Feedback

Comments